PWM to DC Dual DAC LTC2644 interesting...

LTC2644 comes in 8, 10, 12 bit resolution and converts PWM to DC FAST - 10us or so I think. Could be useful - has internal reference voltage. There's a quad part too LTC2645 . Just another possibility in our designs.... so many chips so little time...

There is one very interesting sentence:
"The DAC outputs update and settle to 12-bit accuracy within 8μs typically"

That is REALLY FAST! This would be an excellent companion to the Picaxe.

Speaking of which, what is the equivalent bit resolution for the Picaxe's PWM output?

It's a nice part ( on paper) that almost no one stocks and it is quite overpriced IMO at about $5 -$6 ea (US). Only comes in 12 Lead MSOP (~5mm x 5mm).

But you will get that famous Linear Tech support, that is if you buy at least 1000 parts, they "might" respond to a support request.

That is the Linear Tech way. They have very interesting and high performance parts.

But they are essentially only interested in the large corporate markets, (the prices in large quantities are not so bad) and their choice of offering newer components mostly in the smaller SMT packages (no DIPS or even SOICs!) means that they are mostly geared towards automated SMT equipment.

On the other hand, I thank them for their LTSpice software.

Tell me about it.

Was on a contract back in 05 and we were interested in an LT part. The first year projected volume was less than 500 units. LT would barely speak to us. Then in 07 we had a project with estimated volume of > 100,000. LT got wind of it and came knocking on the door offering samples,etc. The PM gave them an earful an then kicked them out.

Karma ?

That's the nature of business these days, but it's always been that way to an extent. 20 years ago, when I was specifying components for SONY designs I had parts suppliers calling me and offering me things I won't even mention on a public forum, if I would specify their parts for SONY products. They all wanted to establish a "relationship" with me. A week after I left SONY I couldn't even get a call-back from any of them. The more things change, the more they stay the same. That being said, things are tightening up at almost every company.

Companies like Tek and (HP/Agilent/Keysight) no longer include schematics and service info in their manuals, nor will they even sell you a service manual. Thanks to bean-counters, it has gone from the traditional "techies helping techies" to an absolutely adversarial relationship between supplier and customer. Then after you've bought their high priced $20,000 gear and it needs repair they tell you the product is "no longer supported," and they won't fix it for any price, or provide any info about it even though they no longer sell it, and tell you to buy the new model for twice the price.

I'm just glad I can still provide a company name and still get sample parts from most IC manufacturers. I should see if I can snag a couple of these LTC2644's to play with.

What is the resolution for the Picaxe's PWM output?

I too would like to know - probably changing some of the PWM params and seeing what the duty cycle does would get you there - likely 10bit...

Hi,

If you're only considering the timing of the PWM, then the resolution depends on its period, for example 10-bits (0.1%) at the lowest frequency (count 1024) or 8-bits (0.4%) for a count of 256, etc.. However, this is a case where the "accuracy" can be far higher than the resolution, particularly if using an external resonator. The on-chip hardware (counter) could be accurate to a very small fraction of a microsecond (perhaps even low ppm for the mark-space ratio).

However, if you're considering the PWM as a Digital-Anlogue conversion system, then the effective analogue High and Low levels become relevant. Furthermore, if considering a power-supply type application, the power transistor storage or delay times (bipolar or FET) may become very significant.

Cheers, Alan.

@AC the part mentioned in this post does "instant" output of DC proportionate to mark space ratio of the PWM in.

Thanks Allan,that was the answer I was looking for.

The LTC2644 has available an INTERNAL vRef 10ppm / C so can be not subject to _voltage_ from the PWM - also I presume the PWM mark space in the PICAXE is generated digitally and it wouldn't matter what the clock frequency is unless it's varying rapidly [mark/space being digitaly determined]. Can't say that I'll buy one of these right away but I still think what they've done looks like it may be useful to me...

Noticed that there is a 4 channel conversion unit now LTC2645 selling as low as about $5. Still interesting for fast PWM to V conversion.